Background Prognosis of oesophageal adenocarcinoma is notoriously dism
al. To examine the changing patterns of and treatment strategies for t
his disease, the longitudinal experience of a single institution over
16 years is reported. Methods The study comprised a retrospective revi
ew of 551 consecutive patients operated on for oesophageal cancer betw
een 1979 and 1995, of whom 164 had adenocarcinoma. There were 13 women
and 151 men whose mean age was 61 (range 17-82) years. Results The pr
evalence of adenocarcinoma (P = 0.002), that of early tumours (P less
than or equal to 0.10), and the resectability rate (P less than or equ
al to 0.05) increased throughout the period whereas operative mortalit
y rate decreased (P less than or equal to 0.10). Surgical approach cha
nged without influence on long-term survival. Patients referred from e
ndoscopic surveillance programmes for Barrett's oesophagus (n = 16) ha
d an improved survival rate compared with that of non-surveyed patient
s (P less than or equal to 0.01). Overall 5-year survival after oesoph
agectomy (17 per cent) improved for the period 1991-1995 when compared
with 1979-1982 (P less than or equal to 0.02). Univariate analysis id
entified tumour node metastasis (TNM) stage, number of diseased lymph
nodes, invasion of the oesophageal stump and occurrence of a postopera
tive complication as significant prognostic variables (P less than or
equal to 0.05). Multivariate analysis demonstrated that T stage (P = 0
.0002) was the main independent predictor. Conclusions Recent improvem
ent of results reflects patient selection, increased prevalence of ear
ly tumours, and dramatic reduction of the risks from oesophagectomy. N
ew therapeutic directions should be investigated for locally advanced
tumours.